With development of communications networks, to ensure network performance and service quality, network operators need to keep measuring the network performance and service quality of services carried by networks. Network services are hierarchical. An upper-layer service concept includes or depends on a lower-layer service concept. To calculate a network performance indicator, a massive quantity of bottom-layer inputs need to converge upwards layer by layer to eventually calculate a performance indicator of a top-layer resource. However, for a bottom-layer indicator, data of network elements needs to be calculated first, and a calculation result is output to a second layer from the bottom. After an indicator of the second layer from the bottom is calculated, a result of the indicator is then output to a third layer from the bottom. This process is repeated until an indicator of a top layer is obtained by means of calculation. Consequently, workload of intermediate layers is added, memory resources of the intermediate layers are consumed, and a speed of calculating a performance indicator is reduced to some extent.
Therefore, how to reduce workload of an intermediate layer and rapidly calculate a network performance indicator of a specified resource is an urgent technical problem to be resolved.